Related Experiment Videos
Identifying patients for blood conservation strategies
W A van Klei1, A T Rheineck Leyssius, D E Grobbee
1Department of Perioperative Care, Anaesthesia and Pain Management, University Medical Centre, Utrecht, The Netherlands. wklei@azu.nl
The British Journal of Surgery
|August 23, 2002
Summary
Predicting the need for blood transfusions is improved by considering patient factors beyond just surgery type. A simple algorithm combining operation type and hemoglobin levels effectively identifies high-risk patients.
Area of Science:
- Anesthesiology and Perioperative Medicine
- Transfusion Medicine
- Health Informatics
Background:
- Current methods for estimating homologous blood transfusion needs primarily rely on operation type.
- This study aimed to enhance prediction accuracy by incorporating simple patient characteristics.
Purpose of the Study:
- To quantify the improvement in predicting perioperative homologous blood transfusion by including patient characteristics alongside operation type.
- To develop and validate predictive models for perioperative transfusion needs.
Main Methods:
- Retrospective analysis of 24,509 adult surgical patients.
- Development of univariable and multivariable logistic regression models to predict transfusion.
- Validation of models using internal and external datasets, comparing receiver-operator characteristic (ROC) curve areas.
Main Results:
- Multivariable models significantly outperformed the operation-type-only model (ROC areas 0.95 and 0.94 vs. 0.92).
- A simplified model achieved an ROC area of 0.84 in external validation.
- Patients with preoperative hemoglobin < 13 g/dl undergoing major invasive surgery had the highest transfusion risk (43%).
Conclusions:
- A straightforward algorithm integrating operation type and hemoglobin concentration effectively identifies patients at high risk for perioperative homologous blood transfusion.
- This approach offers improved accuracy in transfusion need estimation.